ABSTRACT The granular layer is the input layer of the cerebellar cortex. It receives information from many brain areas through mossy fibers. It outputs information through granule cell axons to the molecular layer, where they contact other interneurons and Purkinje cells (PCs) [1-3]. Our deep understanding of the anatomy and physiology of the granular layer of the cerebellar cortex has made this structure an ideal model system to study signal processing by local networks. An important consideration when studying granular layer function is that there are different neurochemical phenotypes of granular layer interneurons within a single morphological phenotype of interneuron [4-9]. For instance, unipolar brush cells (UBCs) and Golgi cells can be divided into neurochemical subtypes based on the level of GABA- A, glycine and metabotropic glutamate receptors (mGluRs) [5-7]. This neurochemical diversity is not inconsequential, but it provides each neurochemical subtype with different functional roles. However, little is known about the response of different morphological and neurochemical types of granular layer interneurons in the awake behaving animal. This information is essential for a mechanistic understanding of cerebellar cortex function. Our project is designed to provide this information. Our goal is to begin answering the following questions: What are the computations performed by different classes granular layer interneurons in awake animals? How do these computations help generate PC responses and the behavior? Our experiments will tackle these questions head on. Our model system is the macaque ventral paraflocculus (VPFL), which is a cerebellar structure involved in oculomotor control [10, 11]. We will use single unit recordings, pharmacology, and behavioral measurements. In AIM 1, we will identify different morphological and neurochemical classes of granular layer interneurons and study their response to oculomotor tasks. In AIM 2, we will quantify the effect of pharmacological disrupting granular layer processing on the response of PCs (the sole output of the cerebellar cortex, AIM 2A). We will also quantify the effect of pharmacological disrupting granular layer processing on oculomotor behavior (AIM 2B). Lastly, in AIM 3, we will use the above experimental techniques to test the hypothesis that the VPFL, and more specifically its granular layer, participates in building an internal representation of the eye movement (i.e., a forward model of the eye movement) [12-16]. Our preliminary data suggest that the granular layer of the cerebellar cortex performs spatial and temporal signal transformations necessary for normal motor behavior and that the cerebellar cortex participates in the construction of a forward model of the movement. The experimental approach and preliminary data attest to the tremendous potential of the proposed studies to mechanistically understand granular layer function.